Radio GPT

automation_start_radio.yaml

alias: Réveil intelligent - Démarrer la Musique
description: ""
trigger:
  - platform: time
    at: input_datetime.wake_up_time
condition:
  - condition: state
    entity_id: input_boolean.automation_wake_up
    state: "on"
  - condition: state
    entity_id: input_boolean.reveil_intelligent_musical
    state: "on"
  - condition: state
    entity_id: binary_sensor.home_occupied
    state: "on"
  - condition: or
    conditions:
      - condition: state
        entity_id: binary_sensor.workday_today
        state: "on"
      - condition: state
        entity_id: input_boolean.forcer_le_reveil
        state: "on"
action:
  - service: script.turn_on
    metadata: {}
    data:
      variables:
        voice_provider: home_assistant_cloud
        music_media_player: media_player.chambre_sonos_music_assistant
        radio_media_player: media_player.chambre_sonos_music_assistant
    target:
      entity_id: script.radio
  - wait_for_trigger:
      - platform: state
        entity_id:
          - light.chambre
        to: "off"
    continue_on_timeout: false
  - service: media_player.turn_off
    metadata: {}
    data: {}
    target:
      entity_id:
        - media_player.chambre_sonos_music_assistant
  - service: script.turn_off
    metadata: {}
    data: {}
    target:
      entity_id: script.radio
mode: single

prompt

You are going to generate the text of a fake radio host called "JLo FM"
A Fake radio centered around the smart home of JLo

Even if this prompt is in English, the complete radio is in French, because this houseohld is French.

The host is called Jarvis

In response, I expect a string representing the text of the conversation

Something like this:
"Bonjour à tous et bienvenue sur JLo FM, nous sommes le  5 avril et il est actuellement 8 heures du matin, je suis Jarvis, votre hote pour cette matinale"

I will add the subjects I want the radio to mention after the prompt.
I will mention if it is the first message of the stream or not. If it is the first message, please welcome the listeners to JLo FM and present yourself.
If it is not the first message, I will include the past message(s) you spoke, so that you can keep the context.
I will mention if it is the last message of the stream. If it is the last message, please say good bye the listeners of JLo FM and sign off.

script_radio.yaml

alias: Radio
sequence:
  - sequence:
      - service: media_player.turn_on
        data: {}
        target:
          entity_id: "{{music_media_player}}"
      - service: media_player.volume_set
        data:
          volume_level: 0.1
        target:
          entity_id: "{{music_media_player}}"
      - service: mass.play_media
        target:
          entity_id: "{{music_media_player}}"
        data:
          media_id: ⏰ Réveil
          enqueue: replace
          media_type: playlist
      - service: media_player.shuffle_set
        data:
          shuffle: true
        target:
          entity_id: "{{music_media_player}}"
    alias: Start music
  - wait_for_trigger:
      - platform: state
        entity_id:
          - media_player.chambre_sonos_music_assistant
        to: playing
    timeout:
      hours: 0
      minutes: 0
      seconds: 10
      milliseconds: 0
  - delay:
      hours: 0
      minutes: 1
      seconds: 0
      milliseconds: 0
    enabled: true
  - alias: Greetings and Weather
    sequence:
      - service: weather.get_forecasts
        metadata: {}
        data:
          type: hourly
        target:
          entity_id: weather.floirac
        response_variable: weather_forecast
      - service: conversation.process
        metadata: {}
        data:
          agent_id: conversation.gpt_radio
          text: >-
            - It is the first message of the stream 

            - SUBJECT: TIME AND DATE. The current time: {{ now().strftime("%X")
            }} and the current date: {{now().strftime("%x") }}    

            - SUBJECT: WEATHER. The current weather and the forecast for the
            day: {{weather_forecast}}

            - SUBJECT: CURRENTLY PLAYING. The current song playing:
            {{state_attr(music_media_player, "media_title")}} by
            {{state_attr(music_media_player, "media_artist")}}. Add some
            interesting info about the song or artist, if you know. It is not
            mandatory
        response_variable: welcome_radio_generated_response
      - variables:
          welcome_radio_generated_response_formated: "{{ welcome_radio_generated_response.response.speech.plain.speech}}"
      - choose:
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"home_assistant_cloud\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.home_assistant_cloud
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{welcome_radio_generated_response_formated}}"
                  language: fr-FR
                  options:
                    voice: ClaudeNeural
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"elevenlabs\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.elevenlabs_tts
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{welcome_radio_generated_response_formated}}"
  - delay:
      hours: 0
      minutes: 2
      seconds: 0
      milliseconds: 0
  - alias: Chores
    sequence:
      - service: conversation.process
        metadata: {}
        data:
          agent_id: conversation.gpt_radio
          text: >-
            - This is not the first message of the stream

            - Here is your last message for context:
            {{welcome_radio_generated_response_formated}}

            - SUBJECT: TAKING OUT TRASH. Tell me if I need to take out my black
            bin or not. Here is a  boolean that is true if I need to:
            {{states("input_boolean.poubelle_noire_a_sortir")}}. Tell me if I
            need to take out my recycling green bin or not. Here is a boolean
            that is true if I need to:
            {{states("input_boolean.poubelle_verte_a_sortir")}}  

            - SUBJECT: CAT LITTER. Tell me if I need to clean my cat litter.
            HEre is a boolean that is true if I need to:
            {{states("binary_sensor.is_litter_full")}}

            - SUBJECT: CHABAN DELMAS BRIDGE. The next closing of the bridge will
            happen between
            {{as_local(as_datetime(states('sensor.chaban_delmas_next_closing_start_date')))}}
            and
            {{as_local(as_datetime(states('sensor.chaban_delmas_next_closing_end_date')))}}

            Please note that today's date is {{now().strftime("%x") }} and that
            the current time is {{ now().strftime("%X") }}, mention the closing
            of the bridge only it is going to happen in the next few days (or
            today)
        response_variable: task_radio_generated_response
      - variables:
          task_radio_generated_response_formated: "{{ task_radio_generated_response.response.speech.plain.speech}}"
      - choose:
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"home_assistant_cloud\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.home_assistant_cloud
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{task_radio_generated_response_formated}}"
                  language: fr-FR
                  options:
                    voice: ClaudeNeural
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"elevenlabs\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.elevenlabs_tts
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{task_radio_generated_response_formated}}"
  - delay:
      hours: 0
      minutes: 2
      seconds: 0
      milliseconds: 0
  - alias: Calendar
    sequence:
      - service: calendar.get_events
        metadata: {}
        data:
          duration:
            hours: 24
            minutes: 0
            seconds: 0
        target:
          entity_id: calendar.nabucasa_calendar_personal
        response_variable: calendar_events
      - service: conversation.process
        metadata: {}
        data:
          agent_id: conversation.gpt_radio
          text: >-
            - This is not the first message of the stream

            - Here is your last two messages for context.

            - Message 1: {{welcome_radio_generated_response_formated}}

            - Message 2: {{task_radio_generated_response_formated}}

            - SUBJECT: PROFESSIONAL MEETINGS. The meetings I need to attend at
            work. Here are the events for the next 24 hours:
            {{calendar_events}} 

            Please note that today's date is {{now().strftime("%x") }} and that
            the current time is {{ now().strftime("%X") }}. Ignore all events
            happening tomorrow, I am only interested but today's schedule.

            Please remeber that you are a radio host. The text you generate will
            be spoken, so no URL.

            Please remeber that the household is French, try to translate events
            name as much as possible.

            - This is the last message of the stream.
        response_variable: calendar_radio_generated_response
      - variables:
          calendar_radio_generated_response_formated: "{{calendar_radio_generated_response.response.speech.plain.speech}}"
      - choose:
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"home_assistant_cloud\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.home_assistant_cloud
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{calendar_radio_generated_response_formated}}"
                  language: fr-FR
                  options:
                    voice: ClaudeNeural
          - conditions:
              - condition: template
                value_template: "{{voice_provider == \"elevenlabs\"}}"
            sequence:
              - service: tts.speak
                target:
                  entity_id: tts.elevenlabs_tts
                data:
                  cache: true
                  media_player_entity_id: "{{radio_media_player}}"
                  message: "{{calendar_radio_generated_response_formated}}"
mode: single
icon: mdi:radio
fields:
  voice_provider:
    selector:
      select:
        options:
          - label: ElevenLabs
            value: elevenlabs
          - label: Home Assistant Cloud
            value: home_assistant_cloud
    default: home_assistant_cloud
    required: true
    name: Fournisseur de voix
  music_media_player:
    selector:
      entity:
        multiple: false
        filter:
          - integration: mass
            domain: media_player
    name: Lecteur Multimedia pour la Musique
    default: media_player.music_assistant_chambre
    required: true
  radio_media_player:
    selector:
      entity:
        multiple: false
        filter:
          - domain: media_player
    name: Lecteur Multimedia pour la radio
    default: media_player.sonos_chambre
    required: true